Skip to content
Back to search

gtm-starter-kit

SkillWorks with Claude CodeGitHub

A Claude Code repository for GTM teams. Context templates, skills, workflows, and a fully built-out example — fork it, fill in your context, and run AI-powered GTM tasks from a single prompt.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/karlraf--gtm-starter-kit/badge.svg)](/listings/karlraf--gtm-starter-kit)

HTML

<a href="/listings/karlraf--gtm-starter-kit"><img src="/api/listings/karlraf--gtm-starter-kit/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Step 1: Clone and open

git clone https://github.com/KarlRaf/gtm-starter-kit.git
cd gtm-starter-kit
claude .

---

Step 2: Run setup with your domain (15–30 min)

Read skills/setup/SKILL.md and set up this repo for [your-domain.com]

Claude researches your company — website, Crunchbase, LinkedIn, G2, job postings — and writes every context file immediately from public data. No questions asked upfront. You see the repo populate in real time.

When it's done, it shows you what was filled and what was inferred — then offers a 3-minute refinement pass (5 questions) to replace inferred fields with your actual win patterns, ACV range, and signals. Optional. The repo works either way.

---

Step 3: Run your first skill

Read skills/account-research/SKILL.md and research [company.com]

Your context is already there. Claude knows your ICP, your signals, your personas. The output is a full research brief — not a generic summary.

**Other skills to run next:**

# Score a list against your ICP
Read skills/icp-scoring/SKILL.md and score these companies: [paste list]

# Build a campaign from a signal
Read skills/signal-to-sequence/SKILL.md.
Build a Tier 2 campaign for accounts triggering [signal name]. Target [persona].

All outputs land in `outputs/` — see the naming convention below.

---

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 200 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # GTM Starter Kit
  2. Built by [The Revenue Architects](https://www.the-revenue-architects.com) — a GTM engineering firm based in San Francisco. We work with growth-stage B2B SaaS companies to design and build their revenue systems: signal infrastructure, CRM architecture, data pipelines, AI-powered workflows, and the operational layer that connects them.
  3. This repo is the open-source version of what we build with clients. Clone it, fill in your context files once, and run AI-powered GTM tasks from a single prompt — no rebuilding context every session.
  4. ---
  5. ## How It Works
  6. Claude Code reads `CLAUDE.md` automatically at the start of every session. That file contains your company overview, ICP, signals, personas, and current priorities. Once it's filled in, you can run any task with a one-line prompt — no re-explaining your context each time.
  7. ```
  8. Read skills/account-research/SKILL.md and research stripe.com
  9. ```
  10. Claude reads your ICP, signals, and personas from the repo and produces a full account research brief. You didn't paste any context. It was already there.
  11. ---
  12. ## What's Inside
  13. ```
  14. gtm-starter-kit/
  15. │
  16. ├── CLAUDE.md ← Fill this in first. Claude reads it every session.
  17. │
  18. ├── context/ ← Your GTM institutional knowledge
  19. │ ├── profile.md ← Company overview, product, team, reference customers
  20. │ ├── icp-definition.md ← ICP tiers, filters, anti-ICP, qualification criteria
  21. │ ├── signal-library.md ← Signals with scoring, detection methods, hooks
  22. │ ├── positioning.md ← Value pillars, competitive positioning, what not to say
  23. │ ├── competitor-radar.md ← Battlecards, win/loss patterns
  24. │ └── personas/
  25. │ └── template.md ← Persona template — duplicate for each buyer role
  26. │
  27. ├── sync/ ← Scripts for pulling live data into the repo
  28. │ ├── README.md ← Setup and usage
  29. │ ├── .env.example ← Copy to .env and fill in API keys
  30. │ ├── sync-campaign-results.py ← Pull reply/meeting rates from outbound tool
  31. │ └── sync-signal-performance.py ← Aggregate signal performance across campaigns
  32. │
  33. ├── cargo/ ← Your Cargo workspace as code: plays, models, tools
  34. │ └── README.md ← What goes in, setup, rules that prevent drift
  35. │
  36. ├── clay/ ← One file per Clay table: input, waterfall, prompts, cost
  37. │ └── README.md ← The table template
  38. │
  39. ├── skills/ ← Claude executes these. One-line prompts.
  40. │ ├── setup/SKILL.md ← Run first. Provide domain → repo auto-populated.
  41. │ ├── account-research/SKILL.md ← Deep research on any account before outreach
  42. │ ├── signal-to-sequence/SKILL.md ← Turn a signal into a live outbound campaign
  43. │ ├── icp-scoring/SKILL.md ← Score accounts against your ICP, assign tiers
  44. │ └── weekly-update/SKILL.md ← Keep context current. Run Monday mornings.
  45. │
  46. ├── workflows/ ← How your team operates. Human process docs.
  47. │ ├── enrichment.md ← Data waterfall: free sources → Clay → proprietary
  48. │ ├── signal-routing.md ← Decision tree: signal fires → what happens next
  49. │ └── campaign-build.md ← End-to-end: audience → launch → measurement
  50. │
  51. ├── playbooks/ ← Step-by-step for specific situations
  52. │ ├── new-signal-response.md ← Signal fires: validate → score → research → send
  53. │ └── competitor-switch.md ← Four scenarios: unknown, active eval, intent, locked in
  54. │
  55. ├── outputs/ ← All skill outputs land here
  56. │ ├── scoring/ ← ICP scoring outputs
  57. │ ├── campaigns/ ← Campaign briefs and sequences
  58. │ └── .gitkeep
  59. │
  60. └── examples/
  61. └── sample-company/ ← Relay — a fully built-out example
  62. ├── CLAUDE.md
  63. ├── context/
  64. │ ├── profile.md
  65. │ ├── icp-definition.md
  66. │ ├── signal-library.md ← Has real performance data attached
  67. │ ├── positioning.md ← Value pillars, messaging matrix, what not to say
  68. │ ├── competitor-radar.md
  69. │ └── personas/
  70. │ └── head-of-revops.md
  71. └── outputs/
  72. ├── weekly-log.md ← Change history from weekly-update skill runs
  73. ├── 2024-03-15-research-northstar-analytics.md ← Sample research brief
  74. └── campaigns/
  75. └── 2024-03-01-series-b-revops-tier2/
  76. ├── brief.md ← Campaign brief with results tracking
  77. └── sequences.md ← Full email sequences, live copy
  78. ```
  79. **Skills vs. Workflows — what's the difference?**
  80. - **Skills** are for Claude to execute. You run them with a prompt and Claude produces an output. This includes the weekly-update skill, which keeps the repo current rather than producing a deliverable.
  81. - **Workflows** document how your team operates — the decision trees and processes that connect your tools. They're reference documents for humans, not execution instructions for Claude.
  82. ---
  83. ## Getting Started in 3 Steps
  84. ### Step 1: Clone and open
  85. ```bash
  86. git clone https://github.com/KarlRaf/gtm-starter-kit.git
  87. cd gtm-starter-kit
  88. claude .
  89. ```
  90. ---
  91. ### Step 2: Run setup with your domain (15–30 min)
  92. ```
  93. Read skills/setup/SKILL.md and set up this repo for [your-domain.com]
  94. ```
  95. Claude researches your company — website, Crunchbase, LinkedIn, G2, job postings — and writes every context file immediately from public data. No questions asked upfront. You see the repo populate in real time.
  96. When it's done, it shows you what was filled and what was inferred — then offers a 3-minute refinement pass (5 questions) to replace inferred fields with your actual win patterns, ACV range, and signals. Optional. The repo works either way.
  97. ---
  98. ### Step 3: Run your first skill
  99. ```
  100. Read skills/account-research/SKILL.md and research [company.com]
  101. ```
  102. Your context is already there. Claude knows your ICP, your signals, your personas. The output is a full research brief — not a generic summary.
  103. **Other skills to run next:**
  104. ```
  105. # Score a list against your ICP
  106. Read skills/icp-scoring/SKILL.md and score these companies: [paste list]
  107. # Build a campaign from a signal
  108. Read skills/signal-to-sequence/SKILL.md.
  109. Build a Tier 2 campaign for accounts triggering [signal name]. Target [persona].
  110. ```
  111. All outputs land in `outputs/` — see the naming convention below.
  112. ---
  113. ## Output Naming Convention
  114. ```
  115. outputs/YYYY-MM-DD-[type]-[name].md
  116. Examples:
  117. outputs/2024-03-15-research-northstar-analytics.md
  118. outputs/2024-03-01-scoring-q1-prospect-list.md
  119. outputs/campaigns/2024-03-01-series-b-revops-tier2/brief.md
  120. ```
  121. ---
  122. ## Keeping It Current
  123. Run the weekly-update skill every Monday morning:
  124. ```
  125. Read skills/weekly-update/SKILL.md and run the weekly context update.
  126. ```
  127. Claude reads the repo, identifies what's stale, drafts every section that changed, and asks you to fill in the parts it can't know. 10 minutes instead of 45.
  128. Refresh cadence for the rest:
  129. | File | Cadence | Trigger |
  130. |------|---------|---------|
  131. | `CLAUDE.md` priorities | Weekly | Run the weekly-update skill every Monday morning |
  132. | Signal performance log | Every 14 days (minimum) | After each campaign review; more often when campaigns are active |
  133. | `context/competitor-radar.md` | Monthly routine + within 48h of any competitive win or loss | Don't wait for the monthly review after a competitive loss |
  134. | `context/icp-definition.md` | Quarterly | Add an entry to the ICP evolution log; re-score the account list after any change |
  135. | `context/profile.md` | Quarterly | Update after funding rounds, product launches, GTM team changes |
  136. | `context/personas/` | Semi-annually | Review after 3+ similar objections from the same persona suggest the profile is stale |
  137. | `context/signal-library.md` | Ongoing (performance) + quarterly (structural) | Performance log updates are continuous; add or retire signals quarterly based on data |
  138. | `context/positioning.md` | Semi-annually + after major competitive shift | Positioning should be stable; update when the market changes, not on a whim |
  139. | Signal decay re-scoring | Weekly | Run as part of weekly-update; flag accounts where aging signals changed the tier |
  140. ---
  141. ## What Not to Put in This Repo
  142. - **CRM data or contact lists** — never commit customer or prospect data to git
  143. - **API keys or credentials** — use environment variables, never hardcode
  144. - **Raw meeting transcripts** — summarize and synthesize into the relevant context file
  145. - **Pricing** — keep commercial terms out of the repo
  146. ---
  147. ## Example
  148. `examples/sample-company/` is a fully built-out version of the repository for Relay, a fictional workflow automation platform. Every context file is populated. Two real sample outputs are included: an account research brief and a live campaign with email sequences and performance tracking.
  149. Use it as a reference when filling in your own files. The signal library in particular (`examples/sample-company/context/signal-library.md`) shows what a mature, data-backed signal library looks like — with detection methods, decay logic, performance tracking, and signal combination rules.
  150. ---
  151. ## License
  152. MIT

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.